Brain Scans Predict When Poker Players Will Bluff

Your gray matter doesn't have much of a poker face. Even if your stony expression reveals nothing, two bits of brain tissue behind your ears leap into action when you decide to make a bluff. But it's not just about lying. Your brain is deciding whether your foe is a worthy opponent whose actions should affect your own.

Poker playing taps into a skill called social cognition. This is the mental power that lets us assess what other people are up to, figure out what they're thinking, and anticipate what they'll do next. It comes in handy not just for guessing when an opponent has pocket aces, but for everyday activities like negotiating a four-way stop or remembering which of your friends can't keep secrets.

To find out what goes on in our brains while we're judging different opponents, researchers at Duke University set up a very simple game of poker. Don't expect to see this new variant on ESPN anytime soon. The game is played between two players, each of whom is dealt either a high card or a low card. (Rather than numbers, each card reads simply "H" or "L.") The player who goes first decides whether to bet a set amount of money or to fold right away, ceding the ante to the other player. If the first player bets, the second player must decide whether to call the bet or fold.

The game had to be simple, since the 18 study subjects were playing inside a brain scanner. At the beginning of the experiment, subjects were introduced to Player Two (the same person each time). Inside the scanner, they played eight seven-minute rounds of the poker game by hitting buttons on a screen. But there was a twist: At the beginning of half of those rounds, subjects learned they'd be playing against a computer instead of a human.

Subjects were also reminded before playing that bluffing some of the time—that is, betting with a low card and hoping their opponent folded, leaving them the ante money—was a good strategy. The computer opponent was programmed to call players' bets half the time. The human Player Two had been promised extra payment from the researchers if he, too, called bets close to half the time. The result was that the computer and human opponents played the game the same way, says senior author Scott Huettel.

While this nail-biter of a competition was going on, researchers gathered fMRI data from their subjects. They searched everywhere in the brain for areas that lit up with activity when subjects were bluffing. Certain brain regions that had previously been tied to social cognition showed up, indicating the researchers were on the right track.

But there was only one brain region that distinguished between human and computer opponents, the authors reported last week in Science. When the subjects lying in the scanner decided to bluff against a human component—but not against the computer—an area called the temporal-parietal junction (TPJ) suddenly blazed with activity.

Breaking down their results subject-by-subject, the researchers discovered a final twist. After the poker games, they'd asked each subject: Who was a better opponent, the computer or human? "Eleven of the participants said the human," Huettel says, "and the effect was present in those people, but not in the others." In other words, the TPJ only jumps into action when people are bluffing against a human opponent whom they respect.

The TPJ effect was so pronounced that the researchers could use it to tell the future. Huettel says they created a model using one part of their fMRI data, then returned to the rest of the data to test that model. Based on activity in the TPJ when subjects were shown a low card, the researchers could successfully predict whether that subject would bluff or fold about six seconds later.

"There has been an ongoing debate about whether this region is involved in social cognition," Huettel says. The poker study suggests that we do use our TPJs for social cognition, but only in cases where we're basing our actions on another person's behavior. People who thought the computer was a more formidable opponent might have ignored their opponents altogether while deciding whether to bet or fold.

In a real poker game—where there are 52 possible cards rather than two, and decisions about betting or folding are complicated—the TPJ might be at work more often. The activity that predicted a bluff in this study might simply show players are thinking hard about their opponents' cards in a true game of Hold 'Em. Either way, until someone develops remote brain imaging technology, there's probably no need to purchase cerebral sunglasses.

Carter RM, Bowling DL, Reeck C, & Huettel SA (2012). A distinct role of the temporal-parietal junction in predicting socially guided decisions. Science (New York, N.Y.), 337 (6090), 109-11 PMID: 22767930

Image: Rob Watkins/Paf (Flickr)